Vehicle detection apparatus

Abstract

According to one embodiment, a vehicle detection apparatus includes a line segment extraction unit, candidate creation unit, evaluation unit, and specific part detection unit, and the line segment extraction unit extracts a plurality of line-segment components constituting an image of a vehicle from the image formed by photographing the vehicle. The candidate creation unit carries out polygonal approximation configured to create a closed loop by using a plurality of line-segment components to create a plurality of candidates for an area of a specific part of the vehicle. The evaluation unit carries out a plurality of different evaluations for each of the plurality of candidates. Further, the specific part detection unit detects one of the plurality of candidates as the specific part based on evaluation results of the evaluation unit.

What is claimed is: 
     
         1 . A vehicle detection apparatus comprising:
 a line segment extraction unit configured to extract a plurality of line-segment components constituting an image of a vehicle from an image formed by photographing the vehicle;   a candidate creation unit configured to create a plurality of candidates for an area of a specific part of the vehicle by carrying out polygonal approximation configured to create a closed loop by using the plurality of line-segment components;   an evaluation unit configured to carry out a plurality of different evaluations for each of the plurality of candidates; and   a specific part detection unit configured to detect one of the plurality of candidates as the specific part based on evaluation results of the evaluation unit.   
     
     
         2 . The apparatus according to  claim 1 , wherein
 the line segment extraction unit divides the image formed by photographing the vehicle into areas of each identical color based on data of the image, and extracts boundaries between the areas as line-segment components.   
     
     
         3 . The apparatus according to  claim 1 , wherein
 the line segment extraction unit extracts a plurality of line-segment components constituting the image of the vehicle from each of a plurality of images formed by photographing the vehicle, and arranged consecutively in terms of time, and carries out a forecast of geometric variation concomitant with the movement of the vehicle based on these line-segment components arranged consecutively in terms of time to thereby extract a plurality of line-segment components constituting the image of the vehicle.   
     
     
         4 . The apparatus according to  claim 1 , wherein
 the candidate creation unit comprises   a storage unit configured to store a plurality of patterns indicating shapes of parts close to the specific part of the vehicle, and store a plurality of candidates for the specific part associated with the patterns,   a pattern detection unit configured to detect a pattern similar to the part close to the specific part from the storage unit by carrying out polygonal approximation configured to create a closed loop by using the plurality of line-segment components, and   a candidate detection unit configured to detect a plurality of candidates for the specific part associated with the patterns detected by the pattern detection unit from the storage unit.   
     
     
         5 . The apparatus according to  claim 1 , wherein
 the candidate creation unit creates a plurality of candidates for the area of the specific part of the vehicle by carrying out polygonal approximation configured to create a closed loop by supplementing the plurality of line-segment components.   
     
     
         6 . The apparatus according to  claim 1 , further comprising:
 a coordinate detection unit configured to obtain, based on the shooting time, and the shooting position of an image used by the specific part detection unit to obtain a detection result, and a position detected by the specific part detection unit, coordinates of the position on the real space.   
     
     
         7 . A vehicle detection apparatus comprising:
 a mirror detection unit configured to detect right and left side mirrors from an image formed by photographing a vehicle;   a face detection unit configured to detect a face of a driver from the image;   a handle detection unit configured to detect a handle from the image; and   a specific part detection unit configured to detect a position of a windshield in the image based on a detection result of each of the mirror detection unit, the face detection unit, and the handle detection unit.   
     
     
         8 . The apparatus according to  claim 7 , further comprising a coordinate detection unit configured to obtain, based on the shooting time, and the shooting position of an image used by the specific part detection unit to obtain a detection result, and a position detected by the specific part detection unit, coordinates of the position on the real space. 
     
     
         9 . A vehicle detection apparatus comprising:
 a headlight detection unit configured to detect right and left headlights from an image formed by photographing a vehicle;   a license plate detection unit configured to detect a license plate from the image;   a width presumption unit configured to presume a width of the vehicle based on a detection result of each of the headlight detection unit, and the license plate detection unit;   a contour detection unit configured to detect a contour of the vehicle from a plurality of images which are formed by photographing the vehicle and include the image by extracting a boundary between the vehicle and the background; and   a specific part detection unit configured to detect a position of a windshield in the image based on the width presumed by the width presumption unit, and the contour detected by the contour detection unit.   
     
     
         10 . The apparatus according to  claim 9 , further comprising a coordinate detection unit configured to obtain, based on the shooting time, and the shooting position of an image used by the specific part detection unit to obtain a detection result, and a position detected by the specific part detection unit, coordinates of the position on the real space.